Transcriptional profiling of mouse and human ES cells identifies SLAIN1, a novel stem cell gene

Developmental Biology 293 (2006) 90 – 103 www.elsevier.com/locate/ydbio

Transcriptional profiling of mouse and human ES cells identifies SLAIN1, a novel stem cell gene Claire E. Hirst a , Elizabeth S. Ng a , Lisa Azzola a , Anne K. Voss b , Tim Thomas b , Edouard G. Stanley a , Andrew G. Elefanty a,⁎ a b

Monash Immunology and Stem Cell Laboratories, Monash University, Clayton, VIC 3800, Australia The Walter and Eliza Hall Institute of Medical Research, Royal Parade, Parkville, VIC 3050, Australia Received for publication 16 December 2005; accepted 20 January 2006 Available online 20 March 2006

Abstract We analyzed the transcriptional profiles of differentiating mouse embryonic stem cells (mESCs) and show that embryoid bodies (EBs) sequentially expressed genes associated with the epiblast, primitive streak, mesoderm and endoderm of the developing embryo, validating ESCs as a model system for identifying cohorts of genes marking specific stages of embryogenesis. By comparing the transcriptional profiles of undifferentiated ESCs to those of their differentiated progeny, we identified 503 mESC and 983 hESC genes selectively expressed in undifferentiated ES cells. Over 75% of the mESC genes were expressed in hESC and vice versa, attesting to the underlying similarity of mESCs and hESCs. The expression of a cohort of 68 genes decreased greater than 2-fold during differentiation in both mESCs and hESCs. As well as containing many validated ESC genes such as Oct4 [Pou5f1], Nanog and Nodal, this cohort included an uncharacterised gene (FLJ30046), which we designated SLAIN1/Slain1. Slain1 was expressed at the stem cell and epiblast stages of ESC differentiation and in the epiblast, nervous system, tailbud and somites of the developing mouse embryo. SLAIN1 and its more widely expressed homologue SLAIN2 comprise a new family of structurally unique genes conserved throughout vertebrate evolution. © 2006 Elsevier Inc. All rights reserved. Keywords: Embryonic stem cell; Cell differentiation; Transcriptional profiling; SLAIN1

Introduction Embryonic stem cell lines established from the inner cell mass of preimplantation blastocysts can differentiate into ectoderm, mesoderm and endoderm (Keller, 1995; Reubinoff et al., 2000; Thomson et al., 1998), providing an opportunity to study early stages of mammalian development in vitro. Characterising the genes that regulate the stem cell state in both mESCs and hESCs will facilitate the maintenance of undifferentiated ESCs and their directed differentiation to clinically relevant lineages for cell replacement therapies of the future. Several studies have addressed the hypothesis that stem cells isolated from different sources might express a conserved cohort of ‘stemness’ genes. Although two separate reports ⁎ Corresponding author. Fax: +61 3 9905 0780. E-mail address: [email protected] (A.G. Elefanty). 0012-1606/$ - see front matter © 2006 Elsevier Inc. All rights reserved. doi:10.1016/j.ydbio.2006.01.023

comparing the gene expression profiles of mESCs, neural stem cells (NSCs) and haematopoietic stem cells (HSCs) identified 216 and 283 genes, respectively, only 6 genes were common to both lists (Evsikov and Solter, 2003; Ivanova et al., 2002; Ramalho-Santos et al., 2002). When these data were compared with a subsequent study that identified 385 genes enriched in populations of ESCs, NSCs and retinal stem cells, only one gene was common to the three lists (Fortunel et al., 2003). These findings prompted the authors to question the concept that a universal set of ‘stemness’ genes accounted for the properties of self-renewal and multipotency common to stem cells from different sources (Fortunel et al., 2003). In an attempt to identify genes specific to ESCs, the transcriptional profiles of ESCs were compared to a variety of non-stem cell sources including tumor samples, normal tissues and cell lines. In this manner, anywhere from 124 to 1760 cDNAs more highly expressed in ESCs were identified (Richards et al., 2004; Sperger et al., 2003; Tanaka et al.,

C.E. Hirst et al. / Developmental Biology 293 (2006) 90–103

2002). When gene sequences from cDNA libraries generated from undifferentiated and differentiated hESCs were compared, Brandenberger et al. derived a list of 532 genes that were increased in undifferentiated cells (Brandenberger et al., 2004b). Highlighting the variability between published results, only 9 of this list of 532 genes were common with the ‘stemness’ genes generated in the study of Ramalho-Santos et al. and only 32 with the ‘signature’ genes of Ivanova et al. (Ivanova et al., 2002; Ramalho-Santos et al., 2002). When Bhattacharya and colleagues compared sets of genes enriched in mESCs and hESCs, they found that less than 30% of their 92 hESC ‘stemness’ genes were represented in published lists of mESCs genes (Bhattacharya et al., 2004; Ivanova et al., 2002; Ramalho-Santos et al., 2002; Tanaka et al., 2002). Similarly, a study by Sato et al. compared 918 genes enriched in undifferentiated H1 hESC to published mESC-derived data (Ramalho-Santos et al., 2002) and found that 227/541 (42%) of the genes with mouse orthologues were highly enriched in mESCs, also indicating a limited overlap in the gene expression profiles of hESC and mESC (Sato et al., 2003). In this study, we analyzed the transcriptional profiles of differentiating mESCs and demonstrate the sequential expression of genes associated with embryonic epiblast, primitive streak, mesoderm and endoderm, consistent with the hypothesis that in vitro differentiation of ESCs recapitulates molecular changes associated with normal embryonic development. These data provided a rational framework for our prediction that the expression of genes specific to undifferentiated mESC would decrease by day 4 of differentiation, the time when germ layer genes were activated. Using this criterion, a list of 503 putative mESC stem cell genes was identified and similar transcriptional profiling of undifferentiated and differentiated hESCs led to the identification of an analogous population of 983 hESC putative stem cell genes. Attesting to the underlying similarity of mESCs and hESCs, greater than 75% of the mESC stem cell genes were expressed in undifferentiated hESCs and vice versa. A cohort of 68 of these genes, whose transcription decreased greater that 2fold during differentiation of both mESCs and hESCs, included many previously validated ESC stem cell genes as well as a novel gene, designated SLAIN1 (FLJ30046). In this report, we show that SLAIN1 is expressed at the stem cell and epiblast stages of ESC differentiation and in the epiblast, nervous system, tailbud and somites of the developing mouse embryo. SLAIN1 and its more widely expressed homologue SLAIN2 form a new family of structurally unique vertebrate genes. Materials and methods Culture and differentiation of mESCs and hESCs W9.5 ES cells were cultured as previously described (Barnett and Kontgen, 2001). The Slain1-βgeo gene trapped mESC lines (XG385, XG752 and XH855) were obtained from the NIH sponsored NHLBI-BayGenomics (http:// baygenomics.ucsf.edu/overview/projorg.html) and NCRR-Mutant Mouse Regional Resource Center (http://www.mmrrc.org/). The Slain1-βgeo ES cell lines were cultured under antibiotic selection (150 μg/ml Geneticin) in Knockout™ DMEM supplemented with 15% Knockout™ Serum Replacement, nonessential amino acids, 2 mM Glutamine, 50 units/ml Penicillin, 50 μg/ml Streptomycin (all from Invitrogen), 50 mM 2-mercaptoethanol (Sigma), and

91

1000 units/ml LIF (Chemicon). Both the W9.5 and the Slain1-βgeo ES cell lines were differentiated as described (Ng et al., 2005a). Briefly, mESCs were seeded at 5000 cells/ml in Differentiation Media comprising Iscove's Modified Dulbecco's Medium (Invitrogen) supplemented with 15% FCS in non-adherent dishes for the indicated number of days. Alternatively, cells were differentiated under serum-free conditions in Chemically Defined Medium (Wiles and Johansson, 1999) modified as described (Ng et al., 2005a) in the presence or absence of 10 ng/ml of BMP4 (R&D Systems). The hES2 cell line was maintained on monolayers of irradiated PMEFs as previously described (Reubinoff et al., 2000). hESC differentiation was induced by embryoid body formation in hanging drops in Differentiation Media for 5 days. Embryoid bodies were then plated on gelatin-coated dishes for a further 9 days to allow further differentiation and expansion.

Flow cytometry Embryoid bodies were dissociated using trypsin/EDTA (Invitrogen) supplemented with 1% chicken serum (Hunter). FACS-Gal analysis was performed as described (Elefanty et al., 1999; Fiering et al., 1991). In some experiments, dissociated cells were stained with antibodies directed against Flk1 (directly conjugated to PE: BD Biosciences) or E-cadherin (Zymed) detected with allophycocyanin-conjugated goat anti-rat IgG (BD Biosciences) prior to FACS-Gal analysis. Cells were analyzed using a FACSCalibur running CellQuest software (Becton Dickinson).

GeneChip sample preparation and hybridisation Biotin-labelled cRNA was prepared essentially as specified by the Affymetrix standard protocol. Briefly, total RNA was isolated from cells using QIAShredder and RNeasy mini columns including an in-column DNase I digestion (Qiagen). cDNA was synthesized from 7.5 μg of total RNA using an oligo dT(24) primer containing a T7 RNA polymerase binding site. Biotinylated cRNA was produced using the BioArray High-Yield Transcript Labeling Kit (Enzo Diagnostics). Fifteen micrograms of labelled cRNA was fragmented for 35 min at 94°C in fragmentation buffer before hybridisation. Hybridisation and washing were performed according to Affymetrix recommended protocols.

Data analysis Raw data analysis Scanned GeneChip images were analyzed using MicroArray Suite 5.0 and additional analysis performed using Data Mining Tool 3.0 (Affymetrix). Microarray data are available at the ArrayExpress web site (http://www.ebi.ac. uk/arrayexpress/) via Accession numbers E-MEXP-303 and E-MEXP-304. Identification of differentially expressed genes Duplicate samples of undifferentiated mESCs and d4 mEBs from two independent differentiations were analyzed by pairwise comparisons. Samples of undifferentiated hES2 cells (from passage 19 and 128) were compared to EBs differentiated for 5 days (d5 hEBs) in hanging drops or d5 hEBs that were plated onto gelatin for further 9 days (d14 hEBs). ESCGs were characterised by the following criteria: (1) they were present in both baseline samples, (2) they were decreased in at least 3 of the 4 pairwise comparisons between undifferentiated and differentiated samples, and (3) there was at least a 2-fold decrease in 2 of the 4 pairwise comparisons. Comparison of human and mouse enriched genes Orthologues for ESCGs located on either the Human HG-U133A and B or mouse MG-U74Av2 GeneChips were identified using a combination of the NetAffx website (http://www.netaffx.com) and BLASTn searches (http://www. ncbi.nlm.nih.gov/BLAST/).

Expression analysis Total RNA was extracted from adult mouse organs using the RNeasy Midi Kit according to the manufacturer's instructions (Qiagen). A multi-tissue RNA Northern blot containing 10 μg of total RNA for each tissue was probed with a

92

C.E. Hirst et al. / Developmental Biology 293 (2006) 90–103

random-primed radiolabelled probe (MegaPrime Labelling Kit) spanning exons 4 to 9 of mouse Slain1. Semi-quantitative RT-PCR was performed as previously described (Elefanty et al., 1997). Thirty cycles of amplification were used for all the primer sequences at the annealing temperatures shown in Supplementary Table 4.

Whole mount in situ hybridisation A probe coding for nucleotides 894 to 1848 of the mouse Slain1 cDNA (accession number BC079866) was cloned into pBluescriptKS. Whole mount in situ hybridisation was performed using in vitro transcribed, digoxygeninlabelled cRNA essentially as described previously (Pera and Kessel, 1997).

Results Differentiating mouse ES cells sequentially express developmental stage-specific genes In order to document the repertoire of genes expressed in ESC and their differentiated progeny, we analyzed the transcriptional profiles of mESCs at daily intervals over 6 days of differentiation using the murine U74Av2 Affymetrix GeneChip. The microarray data were validated by semiquantitative RT-PCR analysis of selected genes that marked sequential developmental stages in the embryo, including inner cell mass (Oct4 [Pou5f1], Rex1 [Zfp42], and Gbx2), epiblast (Fgf5), primitive streak (Brachyury [T]), and haematopoietic mesoderm (Flk1 [Kdr] and βH1 globin [Hbb-b1]) (Supplementary Fig. 1). These data correlated well with previously reported patterns of gene expression during ES cell differentiation (Keller et al., 1993; Lacaud et al., 2002; Ng et al., 2005a; Robertson et al., 2000). GeneChip analysis revealed the concerted expression of differentiation stage-specific genes in a series of distinct ‘waves’ (Fig. 1). The ‘signature’ genes that have been chosen to demonstrate this phenomenon have been shown by others to be expressed at the appropriate stages in embryonic development, thus establishing a link between the data generated from in vitro ES differentiation and the embryo. The expression of most validated stem cell genes such as Oct4, FoxD3, Rex1, Gbx2, and Sox2 was high in undifferentiated mESCs and had waned significantly by d4 of differentiation consistent with their expression in the early embryo (see examples in Pelton et al., 2002) (Fig. 1A). The biphasic expression of Nanog, with peaks of expression at d0 and at d3 of differentiation, mirrored its distribution in vivo in the inner cell mass of the preimplantation blastocyst and later in the epiblast adjacent to the elongating primitive streak (Hart et al., 2004). The second wave of genes that peaked in expression at d2–3 included the epiblast marker Fig. 1. Sequential expression of genes during mESC differentiation reflects successive stages of embryological development. Relative transcript level for sets of genes indicative of the developmental stages representing (A) stem cells, (B) epiblast, (C) primitive streak, (D) ventral mesoderm, and (E) endoderm. Relative transcript levels represent the signal strength for each transcript expressed as a percentage of the maximal signal for that gene. (F) Expression profile of single genes representing each of the developmental stages shown in panels A–E demonstrating that the temporal relationship between each stage of ESC differentiation mirrors the sequential phases of embryonic development.

C.E. Hirst et al. / Developmental Biology 293 (2006) 90–103

Fgf5 (Haub and Goldfarb, 1991; Hebert et al., 1991; Pelton et al., 2002) and a number of genes including Retinoid X receptor gamma (Rxrg), Occludin (Ocln), and Claudin 6 and 7 (Cldn6 and 7) (Fig. 1B), which have been associated with epithelialization (Kubota et al., 2001; Turksen and Troy, 2001). At d3–4, we identified a third stage of differentiation that was characterised by a rapid, transient increase in transcription of the primitive streak markers such as Brachyury, Eomesodermin [Eomes], Wnt3, and Fgf8 (see examples in Ciruna and Rossant, 1999; Liu et al., 1999) (Fig. 1C). Consistent with the fact that haematopoietic and endothelial cells are the first mesodermal lineages to emerge from the primitive streak (Kinder et al., 1999), we noted an upregulation in the expression of genes marking ventral mesoderm including Flk1, Gata1-3, Runx1, Scl [Tal1], and βH1 globin in d5 and d6 EBs (see examples in Elefanty et al., 1999; Levanon et al., 2001; Yamaguchi et al., 1993) (Fig. 1D and data not shown). Endodermal genes such as Gata6, Sparc, Keratin 19 (Krtl-19), and Collagen type IV, alpha 5 (Col4a5) were also expressed at this stage (see examples in Holland et al., 1987; Morrisey et al., 1996; Tamai et al., 2000) (Fig. 1E). However, it was not possible to unambiguously classify the type of endoderm formed because many of these genes are expressed in both primitive and in definitive endodermal lineages. The sequential overlapping nature of the differentiation stages was evident when the expression levels for individual genes representing examples of each developmental stage were superimposed on the same graph (Fig. 1F). Stem cell genes are conserved in mouse and human ESCs Examination of global changes in gene transcription during mESC differentiation indicated that extensive changes in gene expression accompanied the establishment of germ layers. Specifically, 723 probe sets decreased and 704 probe sets increased more than 2-fold when d0 and d4 samples were compared in one experiment (Fig. 2). Because most of the validated stem cell genes (Fig. 1A) were significantly down-

93

regulated during this interval, we hypothesized that stem cell genes could be operationally defined as genes expressed in undifferentiated ES cells whose transcription decreased in their differentiated progeny. Therefore, we nominated genes whose expression fell by greater than 2-fold between d0 and d4 as embryonic stem cell genes (ESCGs) (see Materials and methods). Of the 503 genes that fulfilled this criterion, 436 (86.7%) represented named genes while the remaining 67 (13.3%) comprised ESTs, RIKEN clones, and other uncharacterised transcripts (Fig. 3A). We hypothesized that human orthologues of mESCGs should also be expressed in hESCs and should also decrease during hESC differentiation. Orthologues represented on the HG-U133 A and B chips were identified for 89.3% (449) of the mESCGs, comprising 94.5% (412/436) of the named genes and 55.2% (37/ 67) of the ESTs/RIKEN clones. Of these human orthologues, 75.9% (341/449) were expressed in undifferentiated hESCs, testifying to the underlying similarity of ESCs in the two species (Fig. 3A). Expression of 34.3% (117/341) genes decreased during differentiation, although the signal intensity decreased less than 2-fold in most cases. The signal of an additional 43.4% (148/341) genes remained stable, and the expression of 22.3% (76/341) genes increased (Fig. 3A and Supplementary Table 1). A complementary experiment in which the transcriptional profiles of undifferentiated hESCs were compared with their differentiated counterparts identified 983 hESCGs that decreased by greater than 2-fold with differentiation (Fig. 3B). 52.6% (517/983) of the hESCGs had mouse orthologues represented on the MG-U74Av2 chip, and of these orthologues, 79.9% (413/517) were expressed in undifferentiated mESCs (Fig. 3B). Similar to our observations with the mESCGs, 37.3% (154/413) of the hESCGs decreased to some degree during mESC differentiation whilst the remainder either remained unchanged (35.1%, 145/413) or increased (27.6%, 114/413) (Fig. 3B and Supplementary Table 2). Combining the 117 mESCGs with human orthologues whose expression decreased during hESC differentiation (Fig. 3A)

▪

Fig. 2. Extensive changes in gene expression accompanied the establishment of germ layers. Graph of global changes in gene expression showing the total number of probes sets at each day of differentiation whose signal had increased (□) or decreased ( ) greater than 2-fold when compared to d0 mESCs. The greatest rate of change in number of probe sets that conformed to the above criteria occurred between day 3 and 4 which corresponds to a time when cells are undergoing the molecular equivalent of gastrulation.

94

C.E. Hirst et al. / Developmental Biology 293 (2006) 90–103

C.E. Hirst et al. / Developmental Biology 293 (2006) 90–103

with the 154 similarly defined hESCGs (Fig. 3B) yielded a total of 203 genes specific to undifferentiated ESCs (udESCGs) (Fig. 4A). Sixty-eight of these genes (58% of the mESCGs) decreased more than 2-fold with differentiation in both species whilst 86 hESCGs and 49 mESCGs downregulated by lesser amounts in both differentiated mouse and human ESCs, respectively (Fig. 4B and Supplementary Table 3). This ‘core group’ of 68 genes included many with validated roles in maintaining the stem cell properties of ESCs such as Oct4, Rex1, Foxd3, Nanog, and Sox2, as well as other genes that play important roles in early embryonic development such as Gdf3, Nodal, Phc1, and Utf1. Classification of the remaining human and mouse ESCGs revealed that these represented intracellular signalling molecules, metabolic enzymes, transcription factors, and secreted factors (Fig. 4B). The 68 core ESCGs also included three incompletely characterised transcripts encoding hypothetical protein FLJ30046, HCV F-transactivated protein 1, and developmental pluripotency associated 4 (Dppa4). Dppa4 has recently been described as a gene with a similar transcriptional profile to Oct4 (Bortvin et al., 2003) and, along with its human orthologue hypothetical protein FLJ10713, has been identified as an ESC gene in several recent transcriptional profiling experiments (Brandenberger et al., 2004a; Rao et al., 2004; Richards et al., 2004; Sato et al., 2003; Sperger et al., 2003). However, the other genes, FLJ30046 (with its respective mouse orthologue 9630044O09Rik) and LOC401152 (also known as HCV F-transactivated protein 1 and its mouse orthologue 1810037I17Rik), have not been previously identified as potential stem-cell-specific genes. We have focussed on FLJ30046 and its mouse orthologue 9630044O09Rik, which we designated SLAIN1 due to the presence of a C-terminal ‘SLAIN’ amino acid motif. SLAIN1 and SLAIN2 encode related genes conserved in vertebrates Examination of public sequence databases revealed that SLAIN1 was well conserved in vertebrates, with ESTs from mouse and human 84.8% identical at the nucleotide level. Homologues of SLAIN1 were identified in rat, chicken, cow, pig, frog, and zebrafish (Fig. 4A and Supplementary Fig. 2). A second highly related cluster of transcripts including 5033405K12Rik and KIAA1458 in mouse and humans, respectively, were also identified and designated as SLAIN2 (Supplementary Fig. 3). This gene was also highly conserved in vertebrates, suggesting that SLAIN1 and SLAIN2 represented two members of a novel gene family (Fig. 5A). The 8 exons of the mouse Slain1 gene spanned approximately 55 kb of mouse chromosome 14 (Fig. 6). Sequence analysis of ESTs derived from mouse Slain1 mRNA indicated

95

the presence of a splice variant in which exon 3 was omitted, resulting in a 50 amino acid internal deletion that did not alter the reading frame. This variable splicing was confirmed by RTPCR analysis of cDNA from mouse ESCs and adult mouse brain, from which both longer and shorter products were obtained (data not shown). The shorter human SLAIN1 ESTs corresponded to exons 4 to 8 of the mouse Slain1 gene and mapped to human chromosome 13 in the region syntenic to mouse chromosome 14. Three upstream human SLAIN1 exons were predicted, based on the high degree of similarity between the human and the mouse Slain1 loci, resulting in a complete human SLAIN1 gene also consisting of 8 exons spanning 65 kb of genomic sequence. The sizes of both exons and introns were well conserved between the two species, and all intron–exon boundaries conformed to the AG/GT rule (Fig. 6). A similar genomic organization of human and mouse SLAIN2 genes was observed, with homologues in both species consisting of 8 exons of a similar size and distribution to those found in SLAIN1. The SLAIN2 genes spanned 75 kb of human chromosome 4 and 60 kb of the syntenic region of mouse chromosome 5 (Fig. 6). The high degree of similarity in the nucleotide sequence and genetic structure suggests that SLAIN1 and 2 are members of the same gene family. Conceptual translations of the open reading frames of mouse Slain1 mRNA and the putative full-length transcript of human SLAIN1 predicted proteins of 629 and 640 amino acids, respectively, with 86.3% identity (Fig. 5B). Mouse and human SLAIN2 proteins both predicted to encode proteins of 581 amino acids, were 93.8% identical, and were 37.1% and 36.3% identical to their respective SLAIN1 homologues (Fig. 5B). Analysis of the protein sequence using MotifScan (http:// myhits.isb-sib.ch/cgi-bin/motif_scan) indicated the presence of serine-rich domains in both SLAIN1 and SLAIN2, while SLAIN1 also contained an N-terminal proline-rich domain (Fig. 5B). While proline- and serine-rich domains can mediate protein–protein interactions, at this stage, their functional importance in SLAIN1 and 2 is speculative. SLAIN1 is expressed at the ‘epiblast’ stage of ESC differentiation prior to gastrulation and in the embryonic epiblast and nervous system RT-PCR analysis of Slain1 expression in differentiating mESCs revealed maximal Slain1 mRNA levels on d2, which declined by d6 of mESC differentiation. Similarly, in hESC differentiation, robust expression of SLAIN1 mRNA was observed in undifferentiated hESCs, which also waned during hESC differentiation. These RT-PCR results were consistent with the relative transcript levels observed in the microarray data (Figs. 7A, B). Slain1 was also detected in adult thymus,

Fig. 3. Identification of embryonic stem cell genes (ESCGs) during differentiation of (A) mouse and (B) human ES cells. 503 mouse ESCGs and 983 human ESCGs were identified based on their expression in undifferentiated cells and downregulation by at least 2-fold in differentiated progeny. Orthologues of ESCGs on human and mouse GeneChips were identified and expression in undifferentiated and differentiated cells evaluated. By these analyses, human orthologues of 117 mESCGs were expressed in hESCs and decreased with hESC differentiation and mouse orthologues of 154 hESCGs were expressed in mESCs and decreased with mESC differentiation. See text for further details.

96

C.E. Hirst et al. / Developmental Biology 293 (2006) 90–103

Fig. 4. Embryonic stem cell genes shared between mouse and human ESCs. (A) Summary of the filters used to identify the mouse and human ESCGs described in Figs. 3A and B. The primary filter (1°) identified genes expressed (P) in undifferentiated ESCs (U) that decreased at least 2-fold (dec) in differentiated cells (D). The secondary filter (2°) identified orthologues for these genes in the reciprocal species that were present in undifferentiated cells and downregulated (lo) during differentiation. (B) Venn diagram demonstrating the overlap between the 117 mouse (blue) and 154 human ESCGs (red), demonstrating that 203 genes are enriched in undifferentiated ESCs (udESC genes). This includes 68 genes that were downregulated 2-fold in both mouse and human ESCs (core ESCGs) with 49 and 86 genes that were downregulated to a lesser extent in mouse and human ESCs, respectively. These genes were classified by their GO Biological Process descriptions.

C.E. Hirst et al. / Developmental Biology 293 (2006) 90–103

97

Fig. 5. High degree of homology between mammalian Slain1 and Slain2 orthologues. (A) Dendrogram of mammalian Slain1 and Slain2 genes. Consensus nucleotide sequences from Homo sapiens (Hs), Mus musculus (Mm), Rattus norvegicus (Rn), Bos taurus (Bt), and Sus scrofa (Ss) were aligned and the phylogenetic tree drawn in MegAlign 6.00 (DNASTAR). (B) Slain1 and Slain2 proteins from mouse (Mm) and human (Hs) were aligned using the Clustal W method in MegAlign 6.00 (DNASTAR). Amino acids identical to mouse Slain1 are indicated by white lettering on a black background. An asterisk indicates the SLAIN motif.

Fig. 6. Slain1 and Slain2 gene loci share a conserved intron–exon structure. Nucleotide sequences of human (h) and mouse (m) Slain1 and Slain2 transcripts were aligned to corresponding genomic sequences using MegaBLAST (http://www.ncbi.nlm.nih.gov/blast/megablast.shtml) and intron–exon boundaries confirmed using Spidey (http://www.ncbi.nlm.nih.gov/IEB/Research/Ostell/Spidey). Exon sizes (in nucleotides) are indicated within each box, and approximate intron sizes (in kb) are shown between boxes.

98

C.E. Hirst et al. / Developmental Biology 293 (2006) 90–103

brain, testis, and lung by Northern blot and also in bone marrow and kidney by RT-PCR (Figs. 7C and D). RT-PCR analysis of Slain2 mRNA indicated that it was not expressed in a restricted manner (data not shown) and has not been investigated further.

We identified three mESC lines generated in the BayQ Genomics Project (http://baygenomics.ucsf.edu/overview/projorg. html) in which gene trap vectors had integrated into the mouse Slain1 locus. Three clones (XG385, XG752, and XH855) that contained a βgeo fusion gene inserted into intron B of Slain1 (as illustrated Fig. 8A) were obtained from the Mutant Mouse Regional Resource Center (http://www.mmrrc.org/). Flow cytometry was used to determine the percentage of viable cells expressing βgalactosidase (βgal) from the Slain1 locus during mESC differentiation using the FACS-Gal technique (Fiering et al., 1991). Consistent with the RT-PCR analysis of Slain1 transcription (Fig. 7A), approximately 20% of undifferentiated mESCs expressed Slain1-βgal, and this increased to include over 80% of cells at d3 of differentiation, with the slight delay in the peak in Slain1-βgal expression as compared to the peak in Slain1 mRNA at d2, reflecting the longer processing time required for expression of the βgal protein (Fig. 8B). The proportion of βgal-expressing cells diminished rapidly thereafter. A similar expression profile was observed for each of the three Slain1-βgal ESC lines (Fig. 8C). Based on our transcriptional profiling data (Fig. 1B), the pattern of Slain1 mRNA expression suggested that it represented an ‘epiblast stage’ gene. One prediction of this hypothesis was that Slain1 mRNA would be co-expressed with E-cadherin, a marker of stem cells, epiblast, and primitive streak (Burdsal et al., 1993; Ciruna and Rossant, 2001; Ng et al., 2005a). Indeed, all βgal-expressing cells at d3 co-expressed Ecadherin (Fig. 8D). By d4, some βgal-expressing cells were Ecadherin−, either indicating that Slain1 transcription persisted in some epiblast or primitive streak derivatives or reflecting the long half-life of the βgeo reporter gene (Fig. 8D). Analysis of the same cells for expression of the ventral mesoderm marker Flk1 revealed that most Flk1+ cells did not express βgal, indicating that Slain1 was not significantly expressed in ventral mesoderm. In previous studies, it had been observed that mESCs differentiated in serum-free (SF) media defaulted to neurectodermal differentiation and supplementation of the medium with bone morphogenetic protein (BMP) 4 induced expression of primitive streak markers and ventrally patterned mesoderm, leading to blood cell formation (Ng et al., 2005a; Park et al., 2004; Wiles and Johansson, 1999). Therefore, we investigated the expression of βgal in Slain1-βgal ESCs differentiated in SF medium in the absence and presence of BMP4 (Fig. 8E). In the presence of BMP4, a similar profile of βgal expression was observed as in serum-containing cultures, with over 80% of cells expressing βgal at d3 and fewer than 10% βgal-positive by d6. In Slain1-βgal ESCs differentiated in the absence of BMP4, Fig. 7. Slain1 is expressed during ESC differentiation and in selected adult tissues. (A and B) RT-PCR analysis of Slain1 expression during ESC differentiation indicates that Slain1 expression (A) peaked at d2 and d3 of mESC differentiation and (B) was robustly expressed in undifferentiated hESCs, correlating with the relative transcript levels calculated from GeneChip signals for human and mouse Slain1 probe sets. (C and D) Tissue distribution of mouse Slain1 mRNA. (C) Semi-quantitative RT-PCR analysis of Slain1 mRNA in adult mouse tissues compared to HPRT expression. (D) Northern blot analysis of Slain1 mRNA in adult mouse tissues with 28S and 18S RNA as loading control.

C.E. Hirst et al. / Developmental Biology 293 (2006) 90–103

99

Fig. 8. Expression of Slain1 during mESC differentiation. (A) Cartoon of mouse Slain1 gene structure indicating the insertion site of the En2 βgeo cassette in intron B. (B) Time course of βgalactosidase expression in differentiating XG385 Slain1-βgeo gene trapped ESCs as determined by FACS-Gal. The percentage of βgal-positive cells is shown in the lower right of each plot. (C) Graph showing the percentage (mean ± SD) of βgal-positive cells for all three Slain1-βgeo ESC lines throughout differentiation. (D) Correlation of E-cadherin and Flk-1 with βgal expression in differentiating Slain1-βgeo ESCs. (E) Differentiation of Slain1-βgeo ESCs in serumfree (SF) media supplemented with (+) or without (−) BMP4.

however, βgal expression was maintained in over 60% of cells even at d6. Most of these cells no longer expressed E-cadherin (data not shown), suggesting that the βgal-positive cells were no longer at the epiblast stage and that Slain1 expression continued in primitive neurectodermal cells. Consistent with these predictions, whole mount in situ hybridisation of postimplantation mouse embryos demonstrated widespread expression of Slain1 throughout the E6.5–E7.0 embryo (Figs. 9A–D) that was followed by higher levels of expression in the headfold neurectoderm at E7.5 (Figs. 9E, F). As embryonic development continued, Slain1 mRNA was observed in the neural tube and optic vesicles at E8.5 and then at sites of imminent neural tube closure in the midbrain, hindbrain, and tailbud at E9.0–E9.5 and in the dorsal aspects of the somites (Figs. 9G–L).

Discussion We have utilised transcriptional profiling to demonstrate the sequential expression of genes representing stem cell, epiblast, primitive streak, and germ layer stages of development in differentiating mESCs. These data extend prior observations that ESC differentiation represents a valid facsimile for early post-implantation mammalian embryogenesis by providing expression data for a larger number of stage-specific genes than previous studies have examined (Doetschman et al., 1985; Keller et al., 1993; Leahy et al., 1999; Palmqvist et al., 2005; Schmitt et al., 1991; Shen and Leder, 1992; Wiles and Johansson, 1999; Wiles and Keller, 1991). The concordance between the stage-specific expression of genes observed during

100

C.E. Hirst et al. / Developmental Biology 293 (2006) 90–103

Fig. 9. Slain1 is expressed throughout the E6.5–E7 embryo and subsequently in the neural tube, tailbud, and somites of the post-implantation mouse embryo at E8.5–E9.5. Whole mount in situ hybridisation of mouse embryos hybridised with sense (A, E, G) or anti-sense (B–D, F, H–L) Slain1 probes corresponding to bases 894 to 1848 of the mouse Slain1 cDNA (accession number BC079866). At E6.5, Slain1 is expressed throughout the embryo (A–D), at E7.5, there is enhanced expression in the headfolds (E, F), and from E8.5 to E9.5, in the optic vesicle (ov), neural tube (nt), midbrain (mb), hindbrain (hb), tailbud (tb) and the dorsal aspects of the somites (so).

ES cell differentiation and during embryonic development suggests that the mechanisms that regulate and coordinate these processes in vivo are, to a significant degree, operational in vitro. This is an important validation of the ES differentiation system and gives confidence that the in vitro differentiation process is predictable and that findings in vitro are relevant to the intact organism. Our analysis indicated that, by d4 of ES differentiation, substantial changes in the repertoire of expressed genes had occurred, including the downregulation of known stem cell genes (see Figs. 1 and 2). This information was used to devise an operational definition for ESC genes, using the criteria of expression in undifferentiated ESCs followed by downregulation by the time of germ layer formation. In order to more stringently define genes likely to be of broad functional importance in ESCs, we narrowed the inclusion criteria to only include those mESC genes with human orthologues that were also restricted in expression to undifferentiated hESCs and vice versa. This defined a group of 203 genes enriched in undifferentiated ESC genes (udESCGs) that were downregulated 2-fold during differentiation in one species and also downregulated to some degree in the other (117 from mESCs

and 154 from hESCs). This included a core group of 68 ESCGs that demonstrated a greater than 2-fold reduction in expression during both mouse and human ESC differentiations. A number of different studies have attempted to identify genes whose expression defines the stem cell state (Abeyta et al., 2004; Bhattacharya et al., 2004; Brandenberger et al., 2004b; Fortunel et al., 2003; Ivanova et al., 2002; RamalhoSantos et al., 2002; Rao et al., 2004; Richards et al., 2004; Sato et al., 2003; Sperger et al., 2003; Tanaka et al., 2002). However, the lists of postulated stem cell genes generated in different studies contained few members in common (Palmqvist et al., 2005). Therefore, it was not surprising that the 283 ‘signature’ genes described by Ivanova et al. included only 5 of our 203 udESCGs, that only 8 were incorporated in the 216 ‘stemness’ genes of Ramalho-Santos et al., and that no genes were shared between all three studies (Ivanova et al., 2002; Ramalho-Santos et al., 2002). Furthermore, only 3 of the ‘signature’ genes described by Ivanova et al. and none of the ‘stemness’ genes of Ramalho-Santos et al. were included in our 68 core ESCGs. The discrepancies between these gene lists could be contributed to by the multiplicity of tissues compared in some studies, variations in the purity of the stem cell populations analyzed,

C.E. Hirst et al. / Developmental Biology 293 (2006) 90–103

and the possibility that related but not identical genes performed similar functions in different stem cell populations (Burdon et al., 2002; Fortunel et al., 2003; Ivanova et al., 2003). However, an equally plausible conclusion from these studies is that there is no common set of genes whose expression defines stem cell characteristics for all tissue types. We observed the greatest similarity between our data and that reported by Palmqvist et al., who examined transcriptional changes occurring during the first 72 h of mESCs differentiation (Palmqvist et al., 2005). Comparison of our 503 mESCGs with their 187 genes ‘decreased after 72 h of LIF removal’ indicated that nearly 50% (93/187) of these genes were included in our data set. Surprisingly, only 14 genes were shared with our 68 core ESCGs. This reduced overlap was in part due to the fact that the study of Palmqvist et al. omitted a number of wellcharacterised ESC genes such as Oct4, Nanog, and FoxD3 that were downregulated more slowly and thus were missed by their use of a 72 h endpoint (see Fig. 1). However, comparisons of the results of other studies that searched for genes specifically expressed in ESCs revealed a low degree of overlap. Therefore, it was not surprising that only 3/25 ‘most positively significant genes’ in hESC reported by Sperger et al., 6/66 ‘candidate hESC signature genes’ from Golan-Mashiach et al., 4/92 genes ‘common to 6 hESC lines’ identified by Bhattacharya et al., 11/227 ‘genes enriched in both hESCs and mESCs’ from Sato et al., and 11/133 ‘pluripotent specific genes’ described by Rao and colleagues were included in our list of 68 core ESCGs (Bhattacharya et al., 2004; Golan-Mashiach et al., 2005; Sperger et al., 2003). It transpired that Oct4 was the only gene identified by all of these comparisons. It is notable that our study was the only one that directly compared mouse and human ES cells using material cultured, harvested, and processed in the same laboratory, assayed on the same platform type, and analyzed using the same data analysis software. As highlighted in recent literature, variability in the microarray platform, labelling and hybridisation protocols, and data analysis software contributed significantly to the lack of concordance observed between different groups' comparisons of similar samples (Bammler et al., 2005; Irizarry et al., 2005; Larkin et al., 2005). It is reasonable to suggest that these problems contributed to the disparity between data sets reported in other stem-cell-related gene profiling studies. In this context, it is relevant that we observed a greater than 75% concordance between the genes expressed in mESCs and hESCs, indicating that ESCs from these two species may indeed be more similar than other studies have suggested (Bhattacharya et al., 2004; Ginis et al., 2004; Sato et al., 2003). Furthermore, our list of genes included most of the recognized ESC genes including FGF4, FoxD3, Nanog, Nodal, Oct4, Rex1, Sox2, and Utf1 (as reviewed in (Rao, 2004) and also encompassed other genes with predicted roles in maintaining ESC pluripotency such as SFRP1, a modulator of Wnt signalling, LeftB, a member of the TGFβ signalling pathway, and DNMT3a, a gene required for the establishment and maintenance of methylation patterns in mESCs (Chen et al., 2003). This is the first study to show that this cohort of stem cell

101

genes is downregulated in a similar manner during both hESC and mESC differentiation, consistent with the similar patterns of in vitro differentiation of mESC and hESC that we have observed (Ng et al., 2005a, 2005b). Our core list of 68 ESCGs included three incompletely characterised transcripts encoding hypothetical protein FLJ30046, LOC401152 (gene designation HCV F-transactivated protein 1), and developmental pluripotency associated 4 (Dppa4). Dppa4 was recently identified along with Dppa2, Dppa3 (also known as Stella or PCG7), and Dppa5 (Esg1) in a screen for genes with similar transcriptional profiles to Oct4 (Bortvin et al., 2003). In this study, we focussed our analysis on FLJ30046 (9630044O09Rik), a gene that we subsequently designated SLAIN1 (Slain1). The presence of serine- and proline-rich domains within the protein sequence suggests that Slain1 may interact with other proteins, but the nature of these interactions is not yet clear. In vitro differentiation of Slain1-βgeo gene trapped mESC lines demonstrated a peak of Slain1 expression at d2–3 (epiblast stage), in keeping with the microarray and PCR data. Further analysis of βgalactosidase activity in these lines indicated that Slain1 expression was downregulated in Flk-1 expressing ventral mesoderm but persisted in ESCs differentiating towards neurectodermal in SF medium in the absence of growth factors. The applicability of these results to the early mouse embryo was demonstrated by in situ hybridisation, which demonstrated widespread Slain 1 expression in the pregastrulation embryo and, subsequently, the developing nervous system and somites. In the adult, Slain1 mRNA was found in brain, thymus, testis and lung by Northern blot and also in bone marrow and kidney by RT-PCR, indicating that Slain1 expression was not restricted to ectodermal derivatives. The localisation of Slain1 to mouse chromosome 14 placed the gene within the piebald coat colour-spotting locus that spans an 18 cM interval surrounding the endothelin B receptor (Ednrb) gene. Deletion complementation analysis has defined several critical regions in this locus that delineated distinct phenotypes of respiratory distress syndrome, skeletal patterning defects and spinal cord malformations (Peterson et al., 2002). The Slain1 gene appears to lie within a 1.3 cM deletion region designated as Ednrbs-1Acrg (Welsh and O'Brien, 2000). The Ednrbs-1Acrg mutant mice exhibited predominantly skeletal patterning defects with a range of abnormalities including failed tailbud development, reduced somite numbers, and neural tube defects including underdeveloped and open neural folds (Welsh and O'Brien, 2000). The expression of Slain1 in these regions of the post-gastrulation mouse embryo makes it a possible candidate for the mutant gene. Targeted mutation of the Slain1 locus in mice is now required to clarify its role as candidate for this syndrome. In summary, the inclusion of most of the known ESC genes in our ‘core’ ESCG set vindicates the strategy taken to identify stem-cell-enriched genes and supports the comparison of transcriptional profiles of closely related cell types as a means of identifying genes that mark different stages of development or cellular differentiation.

102

C.E. Hirst et al. / Developmental Biology 293 (2006) 90–103

Acknowledgments This work was supported by the Australian Stem Cell Centre, the National Health and Medical Research Council of Australia, ES Cell International and the Juvenile Diabetes Research Foundation. A.G.E. is a NHMRC Senior Research Fellow. Appendix A. Supplementary data Supplementary data associated with this article can be found in the online version at doi:10.1016/j.ydbio.2006.01.023. References Abeyta, M.J., Clark, A.T., Rodriguez, R.T., Bodnar, M.S., Pera, R.A., Firpo, M.T., 2004. Unique gene expression signatures of independently-derived human embryonic stem cell lines. Hum. Mol. Genet. 13, 601–608. Bammler, T., Beyer, R.P., Bhattacharya, S., Boorman, G.A., Boyles, A., Bradford, B.U., Bumgarner, R.E., Bushel, P.R., Chaturvedi, K., Choi, D., Cunningham, M.L., Deng, S., Dressman, H.K., Fannin, R.D., Farin, F.M., Freedman, J.H., Fry, R.C., Harper, A., Humble, M.C., Hurban, P., Kavanagh, T.J., Kaufmann, W.K., Kerr, K.F., Jing, L., Lapidus, J.A., Lasarev, M.R., Li, J., Li, Y.J., Lobenhofer, E.K., Lu, X., Malek, R.L., Milton, S., Nagalla, S.R., O'Malley, J.P., Palmer, V.S., Pattee, P., Paules, R.S., Perou, C.M., Phillips, K., Qin, L.X., Qiu, Y., Quigley, S.D., Rodland, M., Rusyn, I., Samson, L.D., Schwartz, D.A., Shi, Y., Shin, J.L., Sieber, S.O., Slifer, S., Speer, M.C., Spencer, P.S., Sproles, D.I., Swenberg, J.A., Suk, W.A., Sullivan, R.C., Tian, R., Tennant, R.W., Todd, S.A., Tucker, C.J., Van Houten, B., Weis, B.K., Xuan, S., Zarbl, H., 2005. Standardizing global gene expression analysis between laboratories and across platforms. Nat. Methods 2 (5), 329–330. Barnett, L.D., Kontgen, F., 2001. Gene targeting in a centralized facility. Methods Mol. Biol. 158, 65–82. Bhattacharya, B., Miura, T., Brandenberger, R., Mejido, J., Luo, Y., Yang, A.X., Joshi, B.H., Ginis, I., Thies, R.S., Amit, M., Lyons, I., Condie, B.G., Itskovitz-Eldor, J., Rao, M.S., Puri, R.K., 2004. Gene expression in human embryonic stem cell lines: unique molecular signature. Blood 103, 2956–2964. Bortvin, A., Eggan, K., Skaletsky, H., Akutsu, H., Berry, D.L., Yanagimachi, R., Page, D.C., Jaenisch, R., 2003. Incomplete reactivation of Oct4-related genes in mouse embryos cloned from somatic nuclei. Development 130, 1673–1680. Brandenberger, R., Khrebtukova, I., Thies, R.S., Miura, T., Jingli, C., Puri, R., Vasicek, T., Lebkowski, J., Rao, M., 2004a. MPSS profiling of human embryonic stem cells. BMC Dev. Biol. 4, 10. Brandenberger, R., Wei, H., Zhang, S., Lei, S., Murage, J., Fisk, G.J., Li, Y., Xu, C., Fang, R., Guegler, K., Rao, M.S., Mandalam, R., Lebkowski, J., Stanton, L.W., 2004b. Transcriptome characterization elucidates signaling networks that control human ES cell growth and differentiation. Nat. Biotechnol. 22, 707–716. Burdon, T., Smith, A., Savatier, P., 2002. Signalling, cell cycle and pluripotency in embryonic stem cells. Trends Cell Biol. 12, 432–438. Burdsal, C.A., Damsky, C.H., Pedersen, R.A., 1993. The role of E-cadherin and integrins in mesoderm differentiation and migration at the mammalian primitive streak. Development 118, 829–844. Chen, T., Ueda, Y., Dodge, J.E., Wang, Z., Li, E., 2003. Establishment and maintenance of genomic methylation patterns in mouse embryonic stem cells by Dnmt3a and Dnmt3b. Mol. Cell. Biol. 23, 5594–5605. Ciruna, B.G., Rossant, J., 1999. Expression of the T-box gene eomesodermin during early mouse development. Mech. Dev. 81, 199–203. Ciruna, B., Rossant, J., 2001. FGF signaling regulates mesoderm cell fate specification and morphogenetic movement at the primitive streak. Dev. Cell 1, 37–49. Doetschman, T.C., Eistetter, H., Katz, M., Schmidt, W., Kemler, R., 1985. The in vitro development of blastocyst-derived embryonic stem cell lines:

formation of visceral yolk sac, blood islands and myocardium. J. Embryol. Exp. Morphol. 87, 27–45. Elefanty, A.G., Robb, L., Birner, R., Begley, C.G., 1997. Hematopoietic-specific genes are not induced during in vitro differentiation of scl-null embryonic stem cells. Blood 90, 1435–1447. Elefanty, A.G., Begley, C.G., Hartley, L., Papaevangeliou, B., Robb, L., 1999. SCL expression in the mouse embryo detected with a targeted lacZ reporter gene demonstrates its localization to hematopoietic, vascular, and neural tissues. Blood 94, 3754–3763. Evsikov, A.V., Solter, D., 2003. Comment on “‘Stemness’: transcriptional profiling of embryonic and adult stem cells” and “a stem cell molecular signature”. Science 302, 393 (author reply 393). Fiering, S.N., Roederer, M., Nolan, G.P., Micklem, D.R., Parks, D.R., Herzenberg, L.A., 1991. Improved FACS-Gal: flow cytometric analysis and sorting of viable eukaryotic cells expressing reporter gene constructs. Cytometry 12, 291–301. Fortunel, N.O., Otu, H.H., Ng, H.H., Chen, J., Mu, X., Chevassut, T., Li, X., Joseph, M., Bailey, C., Hatzfeld, J.A., Hatzfeld, A., Usta, F., Vega, V.B., Long, P.M., Libermann, T.A., Lim, B., 2003. Comment on “‘Stemness’: transcriptional profiling of embryonic and adult stem cells” and “a stem cell molecular signature”. Science 302, 393 (author reply 393). Ginis, I., Luo, Y., Miura, T., Thies, S., Brandenberger, R., Gerecht-Nir, S., Amit, M., Hoke, A., Carpenter, M.K., Itskovitz-Eldor, J., Rao, M.S., 2004. Differences between human and mouse embryonic stem cells. Dev. Biol. 269, 360–380. Golan-Mashiach, M., Dazard, J.E., Gerecht-Nir, S., Amariglio, N., Fisher, T., Jacob-Hirsch, J., Bielorai, B., Osenberg, S., Barad, O., Getz, G., Toren, A., Rechavi, G., Itskovitz-Eldor, J., Domany, E., Givol, D., 2005. Design principle of gene expression used by human stem cells: implication for pluripotency. FASEB J. 19, 147–149. Hart, A.H., Hartley, L., Ibrahim, M., Robb, L., 2004. Identification, cloning and expression analysis of the pluripotency promoting Nanog genes in mouse and human. Dev. Dyn. 230, 187–198. Haub, O., Goldfarb, M., 1991. Expression of the fibroblast growth factor-5 gene in the mouse embryo. Development 112, 397–406. Hebert, J.M., Boyle, M., Martin, G.R., 1991. mRNA localization studies suggest that murine FGF-5 plays a role in gastrulation. Development 112, 407–415. Holland, P.W.H., Harper, S.J., McVey, J.H., Hogan, B.L.M., 1987. In vivo expression of mRNA for the Ca++-binding protein SPARC (osteonectin) revealed by in situ hybridization. J. Cell Biol. 105, 473–482. Irizarry, R.A., Warren, D., Spencer, F., Kim, I.F., Biswal, S., Frank, B.C., Gabrielson, E., Garcia, J.G., Geoghegan, J., Germino, G., Griffin, C., Hilmer, S.C., Hoffman, E., Jedlicka, A.E., Kawasaki, E., Martinez-Murillo, F., Morsberger, L., Lee, H., Petersen, D., Quackenbush, J., Scott, A., Wilson, M., Yang, Y., Ye, S.Q., Yu, W., 2005. Multiple-laboratory comparison of microarray platforms. Nat. Methods 2, 345–350. Ivanova, N.B., Dimos, J.T., Schaniel, C., Hackney, J.A., Moore, K.A., Lemischka, I.R., 2002. A stem cell molecular signature. Science 298, 601–604. Ivanova, N.B., Dimos, J.T., Schaniel, C., Hackney, J.A., Moore, K.A., RamalhoSantos, M., Yoon, S., Matsuzaki, Y., Mulligan, R.C., Melton, D.A., Lemischka, I.R., 2003. Response to comments on “‘Stemness’: transcriptional profiling of embryonic and adult stem cells” and “a stem cell molecular signature”. Science 302, 393d. Keller, G.M., 1995. In vitro differentiation of embryonic stem cells. Curr. Opin. Cell Biol. 7, 862–869. Keller, G., Kennedy, M., Papayannopoulou, T., Wiles, M.V., 1993. Hematopoietic commitment during embryonic stem cell differentiation in culture. Mol. Cell. Biol. 13, 473–486. Kinder, S.J., Tsang, T.E., Quinlan, G.A., Hadjantonakis, A.K., Nagy, A., Tam, P. P., 1999. The orderly allocation of mesodermal cells to the extraembryonic structures and the anteroposterior axis during gastrulation of the mouse embryo. Development 126, 4691–4701. Kubota, H., Chiba, H., Takakuwa, Y., Osanai, M., Tobioka, H., Kohama, G., Mori, M., Sawada, N., 2001. Retinoid X receptor alpha and retinoic acid receptor gamma mediate expression of genes encoding tight-junction proteins and barrier function in F9 cells during visceral endodermal differentiation. Exp. Cell Res. 263, 163–172.

C.E. Hirst et al. / Developmental Biology 293 (2006) 90–103 Lacaud, G., Gore, L., Kennedy, M., Kouskoff, V., Kingsley, P., Hogan, C., Carlsson, L., Speck, N., Palis, J., Keller, G., 2002. Runx1 is essential for hematopoietic commitment at the hemangioblast stage of development in vitro. Blood 100, 458–466. Larkin, J.E., Frank, B.C., Gavras, H., Sultana, R., Quackenbush, J., 2005. Independence and reproducibility across microarray platforms. Nat. Methods 2, 337–344. Leahy, A., Xiong, J.W., Kuhnert, F., Stuhlmann, H., 1999. Use of developmental marker genes to define temporal and spatial patterns of differentiation during embryoid body formation. J. Exp. Zool. 284, 67–81. Levanon, D., Brenner, O., Negreanu, V., Bettoun, D., Woolf, E., Eilam, R., Lotem, J., Gat, U., Otto, F., Speck, N., Groner, Y., 2001. Spatial and temporal expression patterns of Runx3 (Aml2) and Runx1 (Aml1) indicates nonredundant functions during mouse embryogenesis. Mech. Dev. 109, 413–417. Liu, P., Wakamiya, M., Shea, M.J., Albrecht, U., Behringer, R.R., Bradley, A., 1999. Requirement for Wnt3 in vertebrate axis formation. Nat. Genet. 22, 361–365. Morrisey, E.E., Ip, H.S., Lu, M.M., Parmacek, M.S., 1996. GATA-6: a zinc finger transcription factor that is expressed in multiple cell lineages derived from lateral mesoderm. Dev. Biol. 177, 309–322. Ng, E.S., Azzola, L., Sourris, K., Robb, L., Stanley, E.G., Elefanty, A.G., 2005a. The primitive streak gene Mixl1 is required for efficient haematopoiesis and BMP4-induced ventral mesoderm patterning in differentiating ES cells. Development 132, 873–884. Ng, E.S., Davis, R.P., Azzola, L., Stanley, E.G., Elefanty, A.G., 2005b. Forced aggregation of defined numbers of human embryonic stem cells into embryoid bodies fosters robust, reproducible hematopoietic differentiation. Blood 106 (5), 1601–1603. Palmqvist, L., Glover, C.H., Hsu, L., Lu, M., Bossen, B., Piret, J.M., Humphries, R.K., Helgason, C.D., 2005. Correlation of murine embryonic stem cell gene expression profiles with functional measures of pluripotency. Stem Cells 23, 663–680. Park, C., Afrikanova, I., Chung, Y.S., Zhang, W.J., Arentson, E., Fong Gh, G., Rosendahl, A., Choi, K., 2004. A hierarchical order of factors in the generation of FLK1- and SCL-expressing hematopoietic and endothelial progenitors from embryonic stem cells. Development 131, 2749–2762. Pelton, T.A., Sharma, S., Schulz, T.C., Rathjen, J., Rathjen, P.D., 2002. Transient pluripotent cell populations during primitive ectoderm formation: correlation of in vivo and in vitro pluripotent cell development. J. Cell Sci. 115, 329–339. Pera, E.M., Kessel, M., 1997. Patterning of the chick forebrain anlage by the prechordal plate. Development 124, 4153–4162. Peterson, K.A., King, B.L., Hagge-Greenberg, A., Roix, J.J., Bult, C.J., O'Brien, T.P., 2002. Functional and comparative genomic analysis of the piebald deletion region of mouse chromosome 14. Genomics 80, 172–184. Ramalho-Santos, M., Yoon, S., Matsuzaki, Y., Mulligan, R.C., Melton, D.A., 2002. “Stemness”: transcriptional profiling of embryonic and adult stem cells. Science 298, 597–600. Rao, M., 2004. Conserved and divergent paths that regulate self-renewal in mouse and human embryonic stem cells. Dev. Biol. 275, 269–286.

103

Rao, R.R., Calhoun, J.D., Qin, X., Rekaya, R., Clark, J.K., Stice, S.L., 2004. Comparative transcriptional profiling of two human embryonic stem cell lines. Biotechnol. Bioeng. 88, 273–286. Reubinoff, B.E., Pera, M.F., Fong, C.Y., Trounson, A., Bongso, A., 2000. Embryonic stem cell lines from human blastocysts: somatic differentiation in vitro. Nat. Biotechnol. 18, 399–404. Richards, M., Tan, S.P., Tan, J.H., Chan, W.K., Bongso, A., 2004. The transcriptome profile of human embryonic stem cells as defined by SAGE. Stem Cells 22, 51–64. Robertson, S.M., Kennedy, M., Shannon, J.M., Keller, G., 2000. A transitional stage in the commitment of mesoderm to hematopoiesis requiring the transcription factor SCL/tal-1. Development 127, 2447–2459. Sato, N., Sanjuan, I.M., Heke, M., Uchida, M., Naef, F., Brivanlou, A.H., 2003. Molecular signature of human embryonic stem cells and its comparison with the mouse. Dev. Biol. 260, 404–413. Schmitt, R.M., Bruyns, E., Snodgrass, H.R., 1991. Hematopoietic development of embryonic stem cells in vitro: cytokine and receptor gene expression. Genes Dev. 5, 728–740. Shen, M.M., Leder, P., 1992. Leukemia inhibitory factor is expressed by the preimplantation uterus and selectively blocks primitive ectoderm formation in vitro. Proc. Natl. Acad. Sci. U. S. A. 89, 8240–8244. Sperger, J.M., Chen, X., Draper, J.S., Antosiewicz, J.E., Chon, C.H., Jones, S.B., Brooks, J.D., Andrews, P.W., Brown, P.O., Thomson, J.A., 2003. Gene expression patterns in human embryonic stem cells and human pluripotent germ cell tumors. Proc. Natl. Acad. Sci. U. S. A. 100, 13350–13355. Tamai, Y., Ishikawa, T., Bosl, M.R., Mori, M., Nozaki, M., Baribault, H., Oshima, R.G., Taketo, M.M., 2000. Cytokeratins 8 and 19 in the mouse placental development. J. Cell Biol. 151, 563–572. Tanaka, T.S., Kunath, T., Kimber, W.L., Jaradat, S.A., Stagg, C.A., Usuda, M., Yokota, T., Niwa, H., Rossant, J., Ko, M.S., 2002. Gene expression profiling of embryo-derived stem cells reveals candidate genes associated with pluripotency and lineage specificity. Genome Res 12, 1921–1928. Thomson, J.A., Itskovitz-Eldor, J., Shapiro, S.S., Waknitz, M.A., Swiergiel, J.J., Marshall, V.S., Jones, J.M., 1998. Embryonic stem cell lines derived from human blastocysts. Science 282, 1145–1147. Turksen, K., Troy, T.C., 2001. Claudin-6: a novel tight junction molecule is developmentally regulated in mouse embryonic epithelium. Dev. Dyn. 222, 292–300. Welsh, I.C., O'Brien, T.P., 2000. Loss of late primitive streak mesoderm and interruption of left–right morphogenesis in the Ednrb(s-1Acrg) mutant mouse. Dev. Biol. 225, 151–168. Wiles, M.V., Johansson, B.M., 1999. Embryonic stem cell development in a chemically defined medium. Exp. Cell Res. 247, 241–248. Wiles, M.V., Keller, G., 1991. Multiple hematopoietic lineages develop from embryonic stem (ES) cells in culture. Development 111, 259–267. Yamaguchi, T.P., Dumont, D.J., Conlon, R.A., Breitman, M.L., Rossant, J., 1993. flk-1, an flt-related receptor tyrosine kinase is an early marker for endothelial cell precursors. Development 118, 489–498.